I am interested in applying to some OR faculty positions this academic cycle and I am hoping to get a better understanding of the field. For context, I come from an (electrical) engineering background, specifically control systems.

Classically, it seems the problems that OR people studied (at least the ones I am most familiar with) were related to decision theory, optimization and scheduling, and queuing theory with the general sentiment that solving these problems would be of interest to a company's operations.

More recently, the lines between OR and other fields have gotten a bit more blurry for me. I am seeing more venues that focus on the intersection between OR, economics (game theory), and computer science. For example, see the recent talks at EC. Also see a new INFORMS conference on the intersection of OR and security.

Question(s): My main question concerns the current trajectory of the OR field and OR departments. Is the focus of OR becoming blurry or has the underlying motivation for the field changed? What's the future for OR: as problems traditionally studied under OR become more central in society, will other fields (e.g. computer science) begin to take over/cannibalize OR departments?


As far as I can tell, industry demand for people who check the "OR" box (whether they are labeled as OR, industrial engineering or management science, or "analytics" with a clear ability to go beyond basic data-molesting) remains strong. As long as there are jobs, there will be majors, and as long as there are majors, there will be departments. (Occasional "cannibalization") may occur during university reorganizations, but I don't anticipate a major wave any time soon.

It is true that some things originally arcane enough to require OR specialists have now been commoditized, either taught to people in other disciplines (such as supply chain management) or baked into commercial software packages that insulate the user from algorithmic details. Even those things (truck routing comes to mind) require specialists when scope or scale grow too much, or when there are "side constraints" or other user-specific elements that the canned programs don't accommodate.

Shortly after I started teaching (management science, in a business school), supply chain took off as a discipline, and to grow its supply chain program my school cannibalized its rather modest management science program. That as approximately 40 years ago, and OR/MS/IE is still chugging along in rather strong shape today (albeit not at my school). So I'm inclined to bet that the next "cannibalization" is a one-off anecdotal event and not a trend.


In my opinion, OR is definitely broadening in scope, but let me try to be a bit more concrete (N.B.: all of this is my opinion, and I hope others will add their views as well):

  • OR is a relatively "old" field, with the classical problems (scheduling, traveling salesman problem etc) studied by many thousands of people. In addition, OR is a very applied field, so in many places you'll find OR solutions just running fine because they are so well established. Therefore, in the classical OR problems, the progress is mostly (but definitely not exclusively) in new variants of these problems (e.g. vehicle routing problem with time windows and fixed pick-up and drop off and uncertainty in time).
  • This has resulted in more and more connections to machine learning and data science in general, see e.g. this recent paper by Bertsimas which highlights a very good example of this connection. So a new faculty in OR should definitely be ready to work interdisciplinarily to have high impact (in my opinion).
  • The recent hype in machine learning has lead to some cannibalization "attempts", where people try to solve OR problems using ML methods. This is not particularly successful though, since a shortest path problem will always be solved best by OR. ML and the like will however enable augmentation of existing algorithms (see also this review paper), c.f. the previous point.
  • In my opinion, the core of OR - smart decision support systems to achieve enhanced performance - will get more and more important, because the more data etc. you have available, the more angles you can take to enhance your decision making. In my opinion, this is one of the most exciting times to be in OR because of the number of possible directions to take.
  • 1
    $\begingroup$ I am curious to see how long the third bullet will stay true. (Side note: What is meant by "solved by OR"?) $\endgroup$
    – Dirk
    Sep 10 '19 at 19:26
  • $\begingroup$ In what sense it will be invalidated? Do you think ML methods will outperform OR? What I mean by “solved by OR” is using traditional optimization solvers/heuristics to solve the problem, rather than constructing ML models that are then used to evaluate and solve the problem instead. $\endgroup$
    – Richard
    Sep 10 '19 at 21:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.